THE ANATOMY OF THE MADBEPOEAEIA. 413

The Anatomy of the Madreporaria: IV.By

O. Herbert Fowler, B.A., Ph.D.,Assistant to the Jodrell Professor of Zoology in University College, London.

With Plates XXXII and XXXIII.

As was pointed out in a former paper (" Anat. Madrep./'iii), the relations between the external body wall and thecorallum of Madreporaria appear to yield a distinctive mor-phological character, and to depend upon the presence orabsence of coenenchyme. In all the genera yet examined inwhich the individual polyps are more or less free and indepen-dent of each other, the body wall is supported upon laminae ofmesogloea and endoderm (generally termed the " peripherallamellae"), which are continuous over the lip of the calyx withthe mesenteries, and bear a constant relation to them. But inall the genera which form coenenchyme, whether belonging tothe Perforata or to the Imperforata, the body wall rests uponthe little spikes or echinulations which stud the surface of thecorallum. From this it would appear that a distinction, betweenforms with and without coenenchyme, is justified by anatomicaldifferences at least as great as those which differentiate Per-forata from Imperforata; but our knowledge of the group ia soslight that any generalisation such as the above is to beaccepted with great caution. Of the 378 genera recognised byProfessor P. Martin Duncan, in his recent " Revision of theMadreporaria" (' Journ. Linn. Soc. Zool./ xvii), we have nowa more or less complete account of the anatomy of only sixteengenera and a few more species, and the very foundations of atrue morphology of the group have yet to be laid. Indeed,

414 G. HERBERT FOWLER.

the interest of some of the forms which I am about to describelies in the fact that they exhibit a certain divergence fromboth the structural types mentioned above. An account of thefollowing species will be found below.

Madracis asperula, p. 414.AmpliiheJia ramea, p. 416.Stephanophyllia formosissima,

p. 418.Spheuotrochus rubesoens, p.421,

Stepbanaria planipora, p. 424.Pocillopora nobilis, p. 425.Seriatopora tenuicornis, sp. n.,

MADRACIS ASPERULA (fig. 1).

For the material for a study of this coral I am indebted tothe courtesy of Mr. John Murray, who has placed in my handsthe remainder of the spirit collection of Madreporaria obtainedduring the voyage of H.M.S. " Challenger."

In systematic position this coral is ranged by Martin Duncanclose to the genus Stylophora, of the anatomy of which Dr. vonKoch, of Darmstadt, has already published an account (' Jen.Zeitschr./ xi). From this, however, Madracis differs in animportant point.

The corallum, which requires here no systematic descrip-tion, branches in digitate lobes, above the general surface ofwhich the thecse of the polyps project slightly. As in Seria-topora, Stylophora, and other such, it is Imperforate; theliving tissues are confined to the external surface, and do notpenetrate into the depth of the colony, the cavity formerlyoccupied by each polyp being shut off by a succession of" tabulae" as growth proceeds.

The septa are eight in number, Madracis thus addinganother to the several exceptions to Milne-Edwards' law,recently described. With this divergence, however, are notcorrelated in Madreporaria such marked structural differencesas characterise the Monauleae, Edwardsiae, &c, which simi-larly diverge from Hexactinian symmetry. The septa areentocoelic only.

The polyps are in the main of Actinian structure; thebody wall clothing the whole colony is composed of the ordi-

THE ANATOMY OF THE MADEEPORAEIA. 415

nary three layers. The t en tac les are apparently both ecto-coelic and entoccelic, and are simple evaginations tipped eachby a single battery of neroatoeysts, as in Seriatopora ("Anat.Madr.," iiij fig. 11). The septa lie each between a pair ofnormal mesenter ies . No differentiation of particular mesen-teries is recognisable, and all extend to about the same depthinto the polyp cavity. The two pairs of directive mesenteriesare well marked, but it is worthy of note that the plane inwhich they lie does not always coincide with the axial-abaxial(dorso-ventral) plane which generally governs the orientationof similar colonies, but is here, in the case of some polyps, atright angles to it.

Madracis is of especial value as affording an intermediatecondition between the two Madreporarian types mentionedabove, in that the body wall is supported on the echinulationsof the coenenchyme only in certain more or less limited aresebetween the polyps, whereas immediately round the calyces arerecognisable such peripheral lamellae as are usually charac-teristic of forms devoid of coenenchyme (fig. 1, B). This isprobably due to the fact that the polyp calyces are slightlyexsert above the general surface of the colony.

Here, therefore, we apparently have a condition morpho-logically intermediate between that of a solitary iraperforatecoral, as, for example, Caryophyllia, and such a ccenenchyma-tous form as Seriatopora. The condition in Caryophyllia(von Koch, ' Morph. Jahrb./ v, fig. 4) and other simple formsis probably the more primitive. The next stage in the seriesseems to be indicated by the existing condition of Lophoheliaprolifera (" Anat. Madrep.," hi, fig. 6), where the polyps andcalyces are free and independent of each other, although acolony is formed by gemmation. Next it would appear that,as coenenchyme was developed (presumably iu order to increasethe solidity and mass of the colony), and the spaces between thevarious polyps of the colony filled gradually with coral frombelow upwards, the peripheral lamellae, as being necessarilyconfined to the immediate neighbourhood of the calyces, wouldbe inadequate to bridge ovsr the intercalated arese of ccenen-

VOL. XXVIII, PART 3 . NEW SER. F F

416 G. HERBERT FOWLER.

chyme, and hence a new means of support for the body wall,namely, by means of the echinulations of the ccenenchymeitself, became necessary. The acquirement of this secondarycondition is apparently represented by Madracis; but as thecalyces are still slightly exsert above the ccenenchyme, theperipheral lamellae are retained in the arese immediately roundthe polyps. Lastly, in Seriatopora (" Anat. Madrep./' in,fig. 9), which may represent the final term of the series, thecalyces are merged in coenenehyme, and no trace of the peri-pheral lamellae is to be found.

As in other ccenenchymatous forms, by fusion of the bodywall with the echinulations, the section of the coelenteron lyingbelow it is broken up into canals which form the only com-munication between the various polyp cavities.

A figure and description of the corallum may be found inIVfilne-Edwards and Haime, ' Aim. Sci. Nat.' (3) xiii, p. 101,pi. iv, fig. 2.

AMPHIHELIA RAMEA (figs. 2, 3).

Some specimens of the live coral were dredged off Lervikin Norway, by Professor E. Ray Lankester, who has been kindenough to entrust them to me for investigation.

Like Lophohelia, this genus forms by gemmation a branchingcolony without any development of ccenenchyme, the separatepolyps being practically independent of each other. Thetheca is thin and imperforate, and marked externally byirregular longitudinal ridges and furrows which are of somemorphological interest. The septa are generally twenty-fourto twenty-eight in number, of which twelve to fourteen aresmallerand ectoccelic, while a similar number are large and entoccelic.The latter unite below into a loose and irregular columella.

The body wall, which is composed of the usual threelayers, forms a continuous sheet over the whole colony, andthis, with a canal system to be shortly described, constitutesthe sole connection between the adult polyps; that is to say,there is no continuity between the mesenteries of any given

THE ANATOMY OF THE MADREPORAEIA. 417

polyp and those of its parent polyp below, though such aconnection might have been expected. The t en t ac l e s areapparently both ectoccelic and entocoelic; but in these, as inother spirit specimens, it is difficult to determine the pointwith certainty. They are covered with batteries of nematocysts,as in Flabellum (" Anat. Madr.,)J i, fig. 9) and others. The mouthdisc and stomodseum are normal, the lumen of the latter beingnearly filled with coils of craspedon or the contorted edge of themesentery. The mesente r ies are generally in twelve tofourteen pairs of normal appearance, and in my specimens boreabundant ova. Unlike the condition recently described inLophobelia (" Anat. Madr.," iii), there are present two pairsof well-developed "directive" mesenteries; the more remark-able as these two genera are most closely allied, and gemma-tion is of a similar character in both cases. All the mesenteriesextend downwards nearly to the parent polyp, but do nottouch it.

The most interesting point in the anatomy of Amphihelialies in the relation existing between the external body wall andthe corallum. For a very short distance round the lip of thecalyx are recognisable, as in other free forms previously de-scribed, the characteristic peripheral lamellae of the mesen-teries (fig. 2, A). Just below the lip, however, the coral growsoutwards at the points to which these lamellae are attached,and the mesogloea lamina of the body wall comes to lie atthese points directly on the broad ridges of coral thus formed(fig. 2, B). A series of canals lined by endoderm is left be-tween these ridges, which are continuous over the lip of thecalyx with the ccelenteron. These canals (fig. 3) are, in themain, parallel to the longer axis of the polyp, but are alsoconnected by transverse branches; they come to lie in aboutthe same radius as the mesenteries, and are generally equal tothem in number, but are very irregular, both in position andformation. Like the body wall which bounds them peripherally,they appear to be continuous with the similar canals of thepolyps above and below.

An examination of the corallum shows that the ridges be-

418 G. HERBERT FOWLER.

tween which the canals lie are not homologous with true costse,since the latter (which are generally regarded as the euds ofthe septa, projecting radially outwards beyond the tlieca) arealso represented round the lip of Amphihelia. They may per-haps bear some relation to the echinulations of coenenchy-matous forms ; but here, as in most questions of Madreporariaumorphology, a close investigation of a large number of alliedforms is necessary for a true explanation of structure.

A figure and description of the corallum may be found inMilne-Edwards and Haime in ' Ann. Sci. Nat/ (3) xiii, p. 86,pi. iv, fig. 3.

STEPHANOPHYLLIA FORMOSISSIMA (figs. 4, 5, 6).

Professor H. N. Moseleyhas kindly handed to me for furtherinvestigation part of a decalcified specimen of this polyp, of•which he has already published a partial account iu his mono-graph of the Deep-Sea Madreporaria obtained by the "Chal-lenger" ("Chall." Rep. ZooL ii, p. 203, pi. xvi). Owing to thefragmentary nature of the material, I am unable to give adetailed account of the general anatomy, but it is still possibleto make out some points of interest.

As is the case in Fungia, and in the embryo Astroides de-scribed by von Koch and Lacaze-Duthiers, the corallum isplanoconvex, resting free and unattached upon its planesurface. The latter constitutes, the theca, and from it thesepta rise perpendicularly upwards. The theca consists ofa large number of concentric and radial trabeculse, and fromthe radial trabeculse rise alternately a mesentery and a septum(fig. 4). The septa are thus both ectocoelic and entocoelic,and, like the concentric trabeculap. from which they spring, liefree in the ccelenteron, not touching the body wall. On theother hand, those radial trabeculse to which the mesenteriesare attached, lie directly on the basal body-wall, and are unitedwith it just as are the ridges and spikes in Amphihelia andccenenchymatous forms.

Stephanophyllia therefore appears to stand in much the samerelation to the family Eupsammidse, with which it is generally

THE ANATOMY OF THE MADRE1?ORARIA.. 419

classed, as does Amphihelia to its family the Oculinidse; thatis to say, it is possible that in the former, as certainly in thelatter, the peripheral lamellae, which are found in such alliedgenera as Rhodopsammia and Lophohelia respectively, havebeen obliterated and functionally replaced by a radial out-growth of coral from their points of attachment. The tra-beculae of coral to which the mesenteries are attached are ofcourse not iu this genus any more than in Amphihelia to beregarded as true costoe, since they bear no relation to the septa,and are a simple outgrowth from the theca.

In several respects StephanophylHa exhibits a resemblanceto the Fungia recently described by Bourne (' Quart. Journ.Micr. Sci.' xxvii). The plano-convex shape of the corallum,and the correlated basal position of the theca are of coursecharacteristic of both genera; besides this, a view of a com-plete mesentery shows that a few synapticulse brace togetherthe septa of this coral, and that between the synapticulse themuscular pleatings of the mesogloea are gathered into strongbundles in a very characteristic manner hitherto described onlyin Fungia (fig. 5). The histological appearance of the cras-pedon is also similar and characteristic in both forms. Onthe whole, StephanophylHa appears to bring Fungia intocloser relations with the Eupsammidae than have been gener-ally allowed. At the same time it is to be remembered thatin Fungia the body wall is, at any rate partially, supported onperipheral lamellae.

The tentacles are ectocoelic as well as entocoelic, and arecovered with batteries of nematocysts. As has been pointedout in other forms, the nematocysts are of two kinds, of whichthe smaller alone are to be found in the tentacles. In thecraspedon, however, the larger kind is of exceptional dimen-sions, measuring as much as "144 mm. by -012 mm.; whilethe smaller form is only "048 in length. A pair of directivemesenteries occurred iu the small fragment used for transversesections.

420 G. HERBERT FOWLER.

O N SOME STRUCTURES PREVIOUSLY DESCRIBED AS CALICOBLASTIC.

The only other point of interest noticed in Stephanophylliais a structure for the firmer attachment of the mesentery tothe corallum, a structure which occurs in many other genera,though a different significance has been hitherto assigned to it.InF labe l lum a labas t rum it is especially well developed, andconsists of a series of processes given off radially into thecorallum by the mesoglcea lamina of the mesentery (fig. 7).These processes are generally laminated, presumably for afurther increase of the surface of attachment, and are directedobliquely downwards. In transverse sections of decalcifiedspecimens of Flabellum and similar forms they often appearas a layer of polygonal bodies, striated as a rule radially, andlying between the mesoglcea and the space occupied beforedecalcification by the corallum (fig. 8). Such bodies were re-cently described by W. L. Sclater (fProc. Zool.Soc./1886) in anaccount of S t ephano t rochus Moseleyanus as calicoblastsor coral-forming cells; but vertical sections, cut parallel to thebroader plane of the mesentery, show that such an explanationis untenable. A calicoblast is ontogenetically derived fromthe basal ectoderm of the embryo, and presumably has thecharacters which we ordinarily associate with a cell. A layerof such calicoblasts, obviously true cells, may be recognised atthe growing points of any coral. The structures in question,however, are merely offsets of the homogeneous mesoglcea ofthe mesentery, and possess neither nucleus nor cell wall; noris the mesoglcea of the Anthozoa, itself a secretion, known toexhibit secretory activity. Again, these processes do not occurat the growing points of the corallum. On the other hand, asthey are to be met with only in the neighbourhood of the linesof attachment of the mesenteries to the corallum, their posi-tion, as well as their shape and lamination, indicate that theirfunction is to provide an increased surface for fixation of themesentery, and a firmer fulcrum for the action of the powerful

1?HE ANATOMY OP THE MADREPORARIA. 42 i

retractor muscles. The reason that their occurrence has beenhitherto overlooked is, that the mesoglcea is stained only whenthe colouring matter, whether carmine or hsematoxylin, isstrong and diffuse; but I have now detected them in Flabellum(2 sp.), Amphihelia, Seriatopora (2 sp.), Pocillopora (2 sp.),Stephanophyllia, and Sphenotrochus : and Bourne has recentlydescribed their occurrence in Mussa (' Quart. Journ. Micr. Sci./xxviii).

Dr. von Heider, of Graz, who was the first to call attentionto the existence of calicoblasts, describes in a recent accountof Dendrophyllia ('Zeitschr.wiss. Zoo!./xliv) a structure whichhe regards as calicoblastic, in terms which apply exactly to atransverse section of these processes. His figures (pi. xxxi,figs. 9, 11) do not, however, clearly show whether the twostructures are identical or not, though it seems probable thatthey are so, and that what he describes as needles of calciumcarbonate inside the " cells/' are the lines of lamination (cf.Bourne, 'Quart . Journ. Micr. Sci.,' xxviii, p. 25). I t shouldbe noticed, on the other hand, that he explicitly statesthat, in some at least of these so-called cells, a nucleus waspresent.

In Stephanophyllia these processes are to be found at thebase of each mesentery, and form a band running down thesides of the radial trabecula to which the mesentery is attached;but their structure is not different from those occurring inFlabellum and other forms, and their position a natural con-sequence of the different position of the theca (fig. 6).

Figures and description of the corallum may be found inMoseley, " Chall." Rep. Zool., ii, p. 198, pis. iv, xiii, xvi.

SPHENOTROCHUS RTJBESCENS (figs. 9—12).

A study of some fragments of this coral, entrusted to me byProfessor Moseley, has produced one or two points of con-siderable interest, though, as with Stephanophyllia, the accountis necessarily incomplete.

The co ra l l um is not unlike that of Flabellum, near which

422 G. HERBERT FOWLER.

itis placed by Martin Duncan (f Journ. Linn. Soc./ xvii). Fromthe latter it differs, in that soft tissues are present on theoutside of the theca. The septa are both ectoccelic andentoccelic, and to both sets correspond true costse, whichextend downwards for some distance over the external surfaceof the corallum.

As was suggested by Moseley ("Chall." Rep. Zool., ii, p. 159),in complete retraction the tentacles are covered over by themouth-disc, which is drawn inwards by a stroug sph inc te rmuscle (fig. 9). This constitutes the first recorded occurrenceof "Rotteken's muscle" among Madreporaria; in longitudinalsection it has the same "diffuse" appearance as that figured bytheBrothers Hertwig,iu S agar t ia parasi t ica ('Jen.Zeitschr./xiii, pi. xix, fig. 18.) The mou th -d i sc itself is much corru-gated, the rugEe consisting of lengthened columnar ectodermcells supported by solid outgrowths of the mesogloea, whichis at this point very thick (fig. 10, A). The ectoderm of theten tac les consists almost entirely of nematocysts, showingbut slight arrangement into batteries. The tentacles are ap-parently entocoelic only. The mesenter ies appear to benormal, but in the fragment used for microscopic sections nodirectives were present. Both retractor (longitudinal) andprotractor (oblique) muscles are exceptionally well developed,the former applied to such arborescent pleatings of mesogloea ashave been described among both Madreporaria and Hexactinise.It is worthy of note that there is so little cohesion between themuscle-fibres and the mesoglcea which affords them support,that they are frequently seen in transverse sections to tearaway with the endoderm, leaving the mesoglcea bare. Overthe whole of the polyp the mesoglcea is unusually stronglydeveloped, and it exhibits at some points, the position of whichappears to imply a more recent secretion than elsewhere, agreat affinity for carmine, and a markedly vacuolated structure.The vacuoles are often empty, but contain, in many cases,brilliantly refractive crystals, the nature of which I was unableto ascertain. Cells are rarely to be seen in the mesoglcea.The processes for attachment of the mesentery to the corallum

THE ANATOMY Ol1 THE MADREPORARIA. 423

(cf. p. 8, supra) are well developed, not ouly on the mesentery,but also on the thickened mesoglcea at the sides of the pseudo-costse (v. infra, and figs. 10, 11, a).

Encapsuled in the mesoglcea lamina of the mesentery and ofits craspedon are numerous ova in various stages of matura-tion. So far as I am aware, all Anthozoan ova as yetdescribed are formed in the endoderm, and then migrateinto the mesoglcea, the latter forming a capsule appressedclosely round them. In Sphenotrochus, however, the ovum issurrounded by a series of deeply staining bodies which liebetween its membrane and the mesogloeal capsule (fig. 12).At first sight these corpuscles appear to consist of extrudedyolk, but in most, if not in all, it is possible to detect a faintnucleus. Nearly triangular in section, their bases are pressedclosely against the capsule, from which they sometimes shrinkaway, allowing their outline to be clearly seen. It is hardlypossible to regard them as other than follicular cells. Theyare not present in the capsule of the smallest ovum observed(•lx'07 mm., diam. nucleus '008 mm.), and their numberincreases up to a certain point with the size of the ovum. Itis probable therefore that they migrate from time to time intothe mesoglcea, as does the ovum itself; and their appearanceindicates that their function may be to supply yolk-materialfor the ripening egg-cell. More deeply staining and largerbodies of irregular outline are found scattered in the endodermof the craspedon, apparently identical with those described inEuphyllia (Bourne, ' Quart. Journ. Micr. Sci./ xxviii, p. 31) ;these may bear some relation to the follicle cells.

The soft tissues external to the theca are also of someinterest. No peripheral lamellae of the mesenteries are recog-nisable, but at the points where they might be expected tooccur are formed outgrowths of corallum comparable to thosein Amphihelia (v. p. 5, supra.), on which the body wallrests (fig. 10, B. a.). The costse corresponding to the ento-coelic septa come also into direct contact with the body-wall(ibid., a 1) , while those of the entoccelic septa are free. At alower level, however, both sets of costse regularly serve for the

424 G. HERBEKT FOWLER.

support of the body wall (fig. 11, a>), but here and there theyfail to touch it (ibid., a2), thus allowing of a cross commu-nication between the longitudinal canals. The number oflongitudinal canals is thus double that of the ectocoelic andentocoelic spaces, since the body wall is in contact both withtrue costse and with pseudocostse (i. e. those which replace theperipheral lamellae). In the single perfect specimen of thecorallum in the British Museum, the pseudocostse are dis-tinctly visible on the upper two thirds of the exterior of thecalyx, but below this are lost in the true costse. Unfortu-nately my fragments did not show the relations of the softtissues in the lowest third.

Figures and a description of the corallum may be found inMoseley, " Chall." Rep. Zool., ii, p. 157, pi. vi.

STEPHANARIA PLANIPORA (fig. 13).

A specimen of a Psammocorid coral, for which I am againindebted to the generosity of Professor Moseley, appears to beidentical with specimens of the above-named species in theBritish Museum. It is not recorded by Quelch (" Chall."Rep. Zool.), and therefore adds another to the " Challenger "species of Reef-corals. For the identification of this andmany other corals, and for much other assistance, I desire toacknowledge my great obligation to Mr. Dendy, late of theBritish Museum.

The boundaries of the various polyps which compose thecolony are so ill defined that it is impossible to decide howmuch of the soft parts seen in a transverse section should bereferred to a particular individual. A glance at a figure of asection cut transversely to the axes of the polyps will showthis (fig. 13). Each polyp possesses a stomatodseum, to whichare attached a certain number of mesenteries, generally sevento ten; between these mesenteries lie entocoelic and ectoccelicsepta, and over each entocoelic (?) septum is placed a tentacleof the simple character already described for Seriatopora,Madracis, &c. (" Anat. Madr.," iii, fig. 11). The muscles,

TfiK ANATOMY OF THE MADREPOKABIA. 425

though obviously present, are not sufficiently developed toadmit of a distinction of the mesenteries into pairs. The con-tinuity of the mesenteries radiating outwards from the polypis constantly broken by synapticulse. The whole of the surfaceof the colony is practically composed of these radiating slips ofmesentery lying between radiating septa, and interrupted bysynapticulae; and, except in the immediate neighbourhood of apolyp, it is often impossible to decide to which individual amesentery is referable, since neither their direction nor mus-culature afford a clue. Contrary to what one might expect(cf. the more or less comparable condition in Pungia; Bourne,' Quart. Journ. Micro. Sci.,J xxvii, pi. xxv, fig. 13), even suchslips of mesentery as are at no point in contact with thestomatodseum often exhibit a filamentar (craspedal) thickening.Even at the uppermost ends of the branches, at a distance ofperhaps a third of an inch from the nearest polyp, the samearrangement of mesenteries between septa (septo-costae) isfound. On these mesenteries the body wall is supported,although in places it appears to rest also on echinulations ofthe septa; as will be seen from the figure, there is practicallyno coenenchyme. Through the inner parts of the colonyramifies a system of canals by which the various polyps are incommunication with each other.

This genus appears to exhibit a distinct degeneration,implied by the low development of the mesenterial filament (aslight lengthening of the ordinary endoderm), and an indi-viduality hardly more marked than in a Poriferan colony.

A figure and description of the corallum may be found inVerrill, ' Trans. Connect. Acad./ i, p. 545, pi. ix, fig. 4.

PoCILLOPORA NOBILIS.

For material of this species I am again indebted to Pro-fesssor Moseley, who obtained it during the voyage of the" Challenger."

The anatomy agrees almost entirely with that of P.brevicornis , recently described ("Auat. Madr.," iii). The

426 G. HERBERT FOWLER.

differentiation of the mesenteries is similar, though perhapshardly so well marked as in the other species, since the mesen-teries numbered 3 and 10 in the diagram of P. brevicornisare often not appreciably longer than 5 and 8. The reticulartissue mentioned as filling in Ser ia topora subula ta andP. brevicornis the spaces from which corallum has beenremoved by decalcification, is in this species very much moreclearly recognisable. It appears to consist of thin strands of(?) protoplasm (? mesoglcea), staining deeply with carmine; itretains the shape of the crystalline ellipsoids of which thecorallum is ultimately composed, and forms an accurate castof the dead internal parts of the colony.

SERIATOPORA TENTJICORNIS, sp. n. (figs. 14, 15).

For a small fragment of a colony of. this species I owe mythanks to Dr. S. J. Hickson, who collected it with other coralsin the Celebes group.

In anatomy this form agrees exactly with the Ser ia toporasubulata , already described (" Anat. Madr.," iii). It appearsto constitute a new species, of which the following are thediagnostic characters:—The branches are thin and finelytapering, exhibiting no tendency to unite in the manner cha-racteristic of most Seriatoporse ; they are very solid in section,and bear four or five rows of calyces. The ccenenchyme is veryregularly ribbed, the ech inu la t ions of which the ribs arecomposed being markedly short and blunt. The calyces pro-ject slightly above the surface of the ccenenchyme ; the septaare generally about ten in number, short, and very irregular.The columel la does not reach to the surface of the calyx; itis spinous, and not so plate-like as in some other species of thegenus.

From S. cal iendrum, to which it is approximate, it differsin that the calyces are farther apart, and project more from theccenenchyme than in that genus; the echinulations are furtherapart, blunter, and shorter; the branches more slender. Thespecimen has been deposited in the British Museum.

THE ANATOMY OF THE HADltEPORAEIA. 427

CONCLUSIONS.

The most important morphological points elucidated by astudy of the species above described are as follows :

1. The apparent law that the body wall, when present, issupported in acoenenchymatous forms upon " peripherallamelUe" of the mesenteries, and in coenenchymatous speciesupon the echinulations of the ccenenchyme, is seen to requiremodification. The two methods of support may coexist in acoenenchymatous form (Madracis), and to a certain extentin an accenenchymatous species (Amphihelia); further, thebody wall in acoenenchymatous species may rest upon pseudo-costse (? homologous with echinulations), either mainly (Am-phihelia) or entirely (Stephanophyllia); or upon both pseudo-costse and true costse (Sphenotrochus). At the same time it isdoubtful whether such exceptions to the law, formulated above,will not eventually prove to be modificatious due to exceptionalconditions, of which we are at present ignorant.

2. The modifications of the mesenteries exhibited by pre-viously described species of Seriatopora and Pocillopora areseen to extend to other species of the same genus.

3. The ultimate attachment of polyp to corallum consists, inmany genera, in a series of laminated offsets of mesoglcea inthe neighbourhood of the mesentery, such structures havingbeen previously described as calicohlastic in function.

4. A sphincter muscle, comparahle to the " Rotteken'smuscle" of Hexactiniae, may occur among Madreporaria (Sphe-notrochus).

5. Follicle cells, which are perhaps immigrants from theendoderm, may surround the ripening ovum as it lies in itsmesoglceal capsule (Sphenotrochus).

6. A case of degradation, tending to obscure the individualityof the polyps, is presented by Stephanaria.

428 G. HERBERT FOWLER.

L I S T OF THE PAPERS CHIEFLY QUOTED.

BOURNE, G. C.—'Quart. Journ. Mior. Sci.,' xxvii (Fungia).

BOURNE, G. C.—' Quart. Journ. Micr. Sci.,' xxviii (Euphyllia, Mussa).

DUNCAN, P. M.—' Journ. Linn. Soc," xvii (revision of genera).

FOWLER, G. H.—" Anat. Madr.," i, 'Quart. Journ. Micr. Sci.,' xxv; ii, ibid.,

xxvii; iii, ibid., xxviii.

VON HEIDER, A.—' Zeitschr. wiss. Zool.,' xliv (Dendrophyllia).

VON KOCH, G.—'Jen. Zeitschr. Naturw.,' xi (Stylophora).

VON KOCH, G.—'Morph. Jahrb.,' v (Caryophyllia).

MOSELEY, H. N.—" Chall." Rep. Zool., ii tStephanophyllia, Sphenotrochus).

SCLATEB, W. L.—'Proc. Zool. Soc.,1 1886 (Stephanotrochus).

EXPLANATION OF PLATES XXXII and XXXIII,

Illustrating Mr. G. Herbert Fowler's paper on "The Anatomyof the Madreporaria," IV.

FIG. 1. MADRACIS ASPERTJIA.—Diagrammatic transverse section through apolyp at three different levels, showing the eight entocoelic septa and eightpairs of mesenteries. A, above the level of the theca; the exsert edges of thesepta only are cut in this section. B, through the theca ; showing the peri-pheral lamellae. C, below the surface of the ccenenchyme. Ttie support ofthe body wall on the echinulations agrees essentially with that of other formspreviously figured (cf. "Anat. Madrep.," ii, fig. 4; or iii, figs. 2, 9). d.," directive" septa, col- Columella. coin. Ccenenchyme. Ectoderm, "blocked"black and white; mesogloea, black; endoderm, grey; corallum, dotted (v.pp. 2 -4 ) .

FIG. 2. AMPHIHELIA RAMEA.—Diagrammatic transverse section throughtwo quarters of a polyp, showing half of the twenty-four septa and of thetwelve pairs of mesenteries. A, through the theca in the region of the ten-tacles, showing the peripheral lamellae. B, below the stomodseum, showingthe external canals between body wall and corallum. Ectoderm, quasi -cellular; endoderm, mesogloea, and corallum as in Fig. 1 (v. pp. 4—6).